The classification of the depth of burn injury using hybrid neural network

Author

Zhao, Sean X. ; Lu, Taiwei

Author_Institution

R&D Div., Phys. Opt. Corp., Torrance, CA, USA

Volume

1

fYear

1995

fDate

20-25 Sep 1995

Firstpage

815

Abstract

This paper reports on a preliminary study of the classification of burn injuries using a neural network enhanced spectrometer system. Each burn injury is classified as superficial or full-thickness. Spectra covering the visible and near infrared range were collected from burn areas and subjected to autoscaling, principal component analysis, signal preprocessing, and pattern recognition. Classification of 56 data sets collected by the University of Washington Burn Center by this method showed 87.5% classification accuracy

Keywords

backpropagation; biomedical measurement; infrared spectroscopy; medical signal processing; multilayer perceptrons; pattern classification; pattern recognition; skin; visible spectroscopy; 56 data sets; University of Washington Burn Center; autoscaling; burn areas; burn injury depth; classification; classification accuracy; full-thickness burn injury; hybrid neural network; near infrared range; neural network enhanced spectrometer system; pattern recognition; principal component analysis; signal preprocessing; spectra; superficial burn injury; visible range; Infrared spectra; Injuries; Multi-layer neural network; Neural networks; Neurons; Pattern recognition; Physical optics; Principal component analysis; Skin; Testing;

fLanguage

English

Publisher

ieee

Conference_Titel

Engineering in Medicine and Biology Society, 1995., IEEE 17th Annual Conference

Conference_Location

Montreal, Que.

Print_ISBN

0-7803-2475-7

Type

conf

DOI

10.1109/IEMBS.1995.575377

Filename

575377